[Gaudi] Fix the OOM issue of Llama-4-Scout-17B-16E-Instruct (#3245)

Signed-off-by: yuanwu <[email protected]>

Author: yuanwu2017

backends/gaudi/server/text_generation_server/models/custom_modeling/flash_llama_modeling.py

@@ -143,12 +143,14 @@ def __init__(
         config.num_key_value_heads = getattr(
             config, "num_key_value_heads", config.num_attention_heads
         )
-        self.rotary_emb = PositionRotaryEmbedding.static(
-            config=config,
-            dim=self.head_size,
-            base=config.rope_theta,
-            device=weights.device,
-        )
+
+        if config.model_type != "llama4_text":
+            self.rotary_emb = PositionRotaryEmbedding.static(
+                config=config,
+                dim=self.head_size,
+                base=config.rope_theta,
+                device=weights.device,
+            )
 
         # `config.attention_multiplier` is used in Granite
         self.softmax_scale = getattr(